1 The Forum Volume 8, Issue Article 12 POLITICAL SCIENCE AND PRACTICAL POLITICS Ten Things Political Scientists Know that You Don t Hans Noel, Georgetown University Recommended Citation: Noel, Hans (2010) "Ten Things Political Scientists Know that You Don t," The Forum: Vol. 8: Iss. 3, Article 12. DOI: /
2 Ten Things Political Scientists Know that You Don t Hans Noel Abstract Many political scientists would like journalists and political practitioners to take political science more seriously, and many are beginning to pay attention. This paper outlines ten things that political science scholarship has learned that are at odds with much of the conventional wisdom of American politics. KEYWORDS: political science scholarship, conventional wisdom Author Notes: Hans Noel is an assistant professor in the Department of Government at Georgetown University. Before going into political science, he worked as a print journalist in Virginia. He would like to thank Simon Jackman, Marc Meredith, Andrew Therriault and Jonathan Woon for useful suggestions and especially Bethany Albertson, Dan Hopkins, Jonathan Ladd, Chloé Yelena Miller, Brendan Nyhan and John Sides for comments on earlier drafts. The author declares all errors and oversimplifications are his own.
3 Noel: Ten Things Political Scientists Know that You Don t 1 If economics is the dismal science, then political science is the dismissed science. The popular experts working on politics, especially journalists, just do not think we have much to say to them. Politics is not a science, they say. Political scientists are too obsessed with minutiae and mathematical models. We do not really know the people involved. We are out of touch. Much of this criticism is valid, or can be. Academics do care about some very obscure stuff. But we also know quite a lot about how things really do work. Politics is not a science, but politics can be studied systematically. And when it is, you learn a few things. That s political science. Knowledge of the things political scientists learn from political science might even be good for political practitioners, journalists and voters. Practitioners and journalists, we might note, may understand a lot of this stuff intuitively. And yet, too often they seem not to understand the things that political scientists do understand, like the fact that day-to-day campaign minutiae do not really matter that much. Of course, practitioners may have strategic incentives for being disingenuous: If the campaign does not matter, that makes campaign consultants (and campaign-trail reporters) less valuable. Just as if polls suggest that Americans favor one s position, one might not dwell on how that suggestion is misleading. Increasingly, however, journalists are noticing. Some of this is due to the hard work of political scientist bloggers. 1 But a growing number of political journalists are also paying more attention. 2 Yet we can say as much as we like about wanting more people to understand our work. We can also do a lot to try to reach out. But in the end, what, exactly, is it we want people to know? What follows are 10 things that political scientists know that it seems many practitioners, pundits, journalists, and otherwise informed citizens do not. They are an idiosyncratic 10, heavy on American politics and, frankly, on my own interests. And since political scientists are, perhaps more so than those in many other disciplines, a diverse and disagreeable lot, some may even dispute some of these claims. (Though see item #10.) Nevertheless, a world where political writers and readers know these findings would be a worthwhile improvement. 1 A small and very incomplete list would have to begin with the team blogging at themonkeycage.org, and include Jonathan Bernstein at plainblogaboutpolitics.blogspot.com, Daniel Drezner at drezner.foreignpolicy.com, Simon Jackman at jackman.stanford.edu/blog/, Jacob Levy at jacobtlevy.blogspot.com, Jim Johnson at politicstheoryphotography.blogspot.com, Seth Masket at enikrising.blogspot.com, Brendan Nyhan at brendan-nyhan.com, Stephen Walt at and the teams at lawyersgunsmoneyblog.com and duckofminerva.blogspot.com. 2 At the 2010 meeting of the American Political Science Association in Washington, D.C., one panel brought five such journalists (Marc Ambinder, Mark Blumenthal, Ezra Klein, Mark Schmitt, and Matthew Yglesias) to the association, and all five said they wanted to more from us.
4 2 The Forum Vol. 8 , No. 3, Article 12 #1. It s The Fundamentals, Stupid The most exciting and visible part of politics is the political campaign. Politicians and their team of strategists, pollsters, and surrogates wage battle for the votes of the public. Slogans are trumpeted. Gaffes are made. Tactics are deployed. And it probably does not matter all that much. At least not as much as the political environment matters. Presidential elections can be forecast with incredible accuracy well before the campaign really begins. In fact, if all you know is the state of the economy, you know pretty well how the incumbent party will do. See, for instance, Figure 1. If you account for a little bit more, like whether the country is at war, how long the president s party has held the office, and which candidate is more ideologically moderate, you can do even better. (Gelman and King, 1993, Vavreck, 2009, Hibbs, 2000, Bartels and Zaller, 2001). Something similar is true for congressional elections, even midterm elections, although the relationships are not as strong, and the evidence is more mixed. Individual races are hard to predict. But in the aggregate, the fundamentals matter again. In midterms, the public s perception of the president has a huge impact. And note that the president s party almost always loses seats in midterm elections. In a down economy, a lot of loss should be expected, even without any reference to the specific policies of the incumbent. We know less about local elections, however, where national economic conditions might be less salient. And many of the fundamentals that matter in the general election do not apply to primary elections (although primaries too have their predictable features. See for instance Mayer, 1996, 2004, 2008, Adkins and Dowdle, 2005, Steger, 2007, Cohen, Karol, Noel, and Zaller, 2008). It is worth poring over Figure 1 a little. Note that both the 2008 election, widely viewed as a decisive repudiation of the overreach of George W. Bush, and the 1980 election, widely viewed as a decisive repudiation of the incompetence of Jimmy Carter, are very close to the regression line. To the extent that those incumbents were responsible for the economic conditions, this is a repudiation of them. But there is no need to appeal to personalities to either of those elections. The vote, after all, is a crude instrument. If voters don t like how things are going, they punish the people who are in power. It may not matter whether the people in power had anything to do with the pain the country is suffering. Christopher Achen and Larry Bartels (2002) find evidence that voters have punished incumbents for droughts, floods, and shark attacks. Still, it is not surprising that people behave this way. Voters are influenced by the world in which they are living more so than any campaign stunt.
5 Noel: Ten Things Political Scientists Know that You Don t 3 Figure 1: Presidential Vote vs. the Economy Presidential Vote vs. Economy, Incumbent's Share of Two-Party Vote % Change in RDI Percent change in Real Disposable Income in last year of incumbent party s term. Bureau of Economic Analysis, September 2010 (www.bea.gov) vs. Incumbent party s share of the vote for the Democrat and Republican parties Leip s Atlas of U.S. Election Results (www.uselectionatlas.org)
6 4 The Forum Vol. 8 , No. 3, Article 12 A lot of politicians understand this. Bill Clinton certainly did: his 1992 campaign mantra was it s the economy, stupid. But if the economy and other fundamentals determine outcomes, why should the campaign have to emphasize it at all? Research by Lynn Vavreck (2009) suggests that candidates can affect how the economy affects the outcome. Candidates who are advantaged by the economy ought to make it the centerpiece of their campaigns, as Clinton did in 1996, and as Al Gore did not in And candidates disadvantaged by it ought to try to change the focus of the race. In other words, campaigns can matter in the way they connect to the fundamentals. Campaigns also remind voters much about their own partisan attachments and inform them about the fundamentals (Gelman and King, 1993). And of course, both parties are campaigning. If one stopped, maybe the other would have an effect. But even the most incompetent campaign does not seem to overwhelm the basic role of the fundamentals in elections, especially in national elections. Certainly campaign stunts and clever sound bites will not do it. #2. The Will of the People is Incredibly Hard to Put Your Finger on How do you know a political commentator is making stuff up? They pretend to know what The American people want, think, will do, or anything else. The first, most obvious, problem is that a majority in a given survey does not represent all of the American people. If 75 percent of respondents say they are for something, this means that 25 percent did not say they were for it. Those 25 percent are Americans, too. But of course, we have a strong belief in majority rule, so perhaps that is not so troublesome. The more general problem is that surveys are incredibly imperfect tools, often interpreted by people who do not know or do not care about how they work. Respondents are influenced by all sorts of artifacts, from question wording and question order to the race and gender of the interviewer. They are also influenced by the political environment in which their answers are given, including the fundamentals (see Item #1). And they are very influences by political elites. And when it comes to policy positions, most people just do not give good answers, often because they are not asked good questions. Simply put, surveys are not simple. Instead, years of research on public opinion has led us to a number of conclusions about how to interpret them. Most People Are Not Very Ideological Liberalism and conservatism are complex belief systems, each drawing on a variety of principles, values and perspectives.
7 Noel: Ten Things Political Scientists Know that You Don t 5 Some people engage those belief systems. Most do not. Some of the earliest work on public opinion (Converse, 1964, Campbell, Converse, Miller, and Stokes, 1960) found that most voters know something about the groups they identify with, and perhaps which party those groups are affiliated with. But liberalism and conservatism, not so much. Some of this may be changing, as the political parties become more ideologically distinct. But it remains true that most voters have an incomplete grasp of what liberal and conservative mean. The exception is the most highly informed, highly partisan voters. The more engaged one is in politics, the more ideological one is. Most People Do Not Have Strong Political Opinions Most people are not political junkies. They have never heard the latest question that has politicos obsessed. They may not know the name of their own congressional representative. If they are aware, they have heard different things, and may not have formed an opinion before a pollster lands on their phone number. None of this means people are stupid, or too apathetic to participate in politics. They just have better things to do with their lives. But if you ask them their opinion on political questions, they will do their best. Maybe they have a few considerations about the subject, and they will grab the first one that pops into their head. When you ask them the next time, they might say something else. But they will say something. And so the poll will have results. Walter Lippmann described this problem in his introduction to Public Opinion (1922), The World Outside and Pictures in Our Heads : The real environment is altogether too big, too complex, and too fleeting for direct acquaintance. Instead, we have impressions that come close to reflecting the world, but are incomplete. A poll saying 55 percent of respondents support some policy cannot reflect just how incomplete. It is not meant to. Most People Take Cues From Parties And Political Leaders The biggest way to get respondents to change their opinion is to tell them what someone else thinks. In survey experiments, when respondents are told that the president or one of the parties supports or opposes a policy, partisans (and even independents, as in item #7 below) are more likely to give the position endorsed by someone they identify with. Democrats support policies associated with Obama and Republicans oppose them. If the respondents are not told the position of party leaders, the division is less pronounced.
8 6 The Forum Vol. 8 , No. 3, Article 12 This also works in a more indirect way. If partisan leaders are consistently taking one position in public, voters who identify with their party will come to agree (Zaller, 1992). If you want to know why so many Americans think Obama is a Muslim and health care reform involved death panels, you might start by looking at what Republican and conservative political leaders are saying. If you want to know why Democrats came to oppose a war in Vietnam that was started by Democratic politicians, look to the signals sent by Democrats over the course of the war. Such cue-taking need not be a bad thing. Keeping track of every policy issue is hard. Following leaders who share your values is a useful short-cut. But the result, once again, is that public opinion is not an entirely independent force constraining political leaders. A lot (although by no means all) of what the public thinks is heavily shaped by what politicians are saying. Let The Polls Be Polls Polls are very good for some things. For instance, they are good when the question is very precise, such as Who will you vote for in the election? It is the why that gets tricky. And many of the problems with public opinion data fade away when we look at aggregate trends. Aggregating polls can wash out some of the uncertainty among voters. Just as we do not grade students on the answer to just one test, we also should not arrive at impressions about the broad preferences of the public on the basis of just one question. This is the difference between most studies in political science that use survey data, versus polls reported in the newspapers. Political scientists try to figure out how respondents are arriving at the answers the give. They work hard to put them into context. Some of this can be done with survey experiments (Ansolabehere and Iyengar, 1995, Iyengar and Kinder, 1987, Sniderman and Carmines, 1997) that manipulate the questions or their context. And some of it can be done by examining many questions over many periods MacKuen, Erikson, and Stimson (1989). #3. The Will of The People May Not Even Exist OK, let us say that the American people do have preferences, even if it is hard to measure them with surveys. We need to aggregate those preferences somehow. We need to let the American people participate in democracy and get collective decisions that are reasonable. That might not be possible.
9 Noel: Ten Things Political Scientists Know that You Don t 7 For instance, if everyone prefers a system of private insurers to a singlepayer health care system, then we should not have single-payer if the private system is possible. We might think of a number of criteria that would apply to any a reasonable method for making collective social choices. Here are five: 1. Collectively Rational: The American people, as a collective, should behave as though they are rational. This means, for example, that if the group prefers Obama to McCain and it prefers McCain to Romney, then it ought to prefer Obama to Romney. Otherwise, we have a cycle that could never be settled. Individual people may sometimes fail to meet this standard of rationality, of course, but we do not want our system of voting to introduce more irrationality. 2. No Restrictions: Whatever rule we have, we cannot say that it only works if citizens only want certain things. Whatever the people prefer, we have to be able to aggregate it. Democracy is not democracy if it only works in places where everyone agrees or where only some questions are on the table. 3. Unanimity: If everyone prefers one alternative to the other, you should not get the other. Majority rule might not be obvious. Maybe we should require a supermajority to change a policy from the status quo. Or maybe we should require a majority in every state, or from every region. We might even require that everyone agree before we change. But if everyone does agree, and we still do not adopt the proposal, then it is hard to call that a democracy. 4. Independence of Irrelevant Alternatives: This is a little tricky, but the idea is that adding new choices should not make you reorder your preferences about old ones. Suppose you and your friends are choosing a place to eat, and you are deciding between an Indian place and a steakhouse. Of those options, you choose the steakhouse, but then someone sees a Tex-Mex place across the street. You might stick with your choice of the steakhouse, or you might go for Tex Mex. But you should not, now faced with the new option of Tex-Mex, suddenly decide on Indian after all. 5. No dictator: No one person s preferences should be able to tip the scales just because of who they are. It is hard to argue that a collective choice rule that does not meet all five of these criteria is really democratic. But social choice theorist Kenneth Arrow
10 8 The Forum Vol. 8 , No. 3, Article 12 (1951) concluded that there is no way to aggregate preferences that satisfy all five criteria. This result is counterintuitive, and perhaps hard to believe. I won t go through the proofs here, but let us just note that it is possible to describe scenarios where violations of these principles will happen for every voting rule you could think of. Maybe those scenarios will never crop up, but since we do not know what everyone thinks, it is hard to be sure that they do not crop up all the time. For instance, simple majority rule violates the first principle above. Suppose people s preferences among Obama, McCain, and Romney as president were as follows: Table 1: Hypothetical Voter Preferences 30% of voters 30% of voters 40% of voters First choice: Obama McCain Romney Second choice: McCain Romney Obama Third choice: Romney Obama McCain If we ask voters to choose between McCain and Romney, McCain will win, because the voters in the first two columns prefer McCain to Romney, and together they make up 70 percent of the electorate. If we ask them to choose between McCain and Obama, Obama wins with the support of the voters in the first and third columns. But Romney would beat Obama in this hypothetical scenario with the votes of those in the second and third columns. This possibility is called Condorcet s paradox (Condorcet, 1785). It may or may not be all that common. The preferences in Table 1 are contrived to make the point. But recall the second principle above: We have to be able to aggregate any preferences. Arrow s Theorem does not mean that democracy is impossible. Only that it might be. Certainly, it means that democracy is a procedure. The institutions we use have consequences. Note that there is nothing here about people not being smart enough, or knowing enough about the issues, or having the time to vote on everything, or anything practical. Just assume away all those nuisances and try to have a true democracy, and it can still break down. And we do see such breakdowns. The election of Woodrow Wilson in 1912 is a perfect example. Theodore Roosevelt and William Howard Taft both ran as Republicans, and they split the Republican vote. So Wilson won, but it is likely that a majority of voters preferred either Roosevelt or Taft, or both, to Wilson. Things broke down even worse in 1824, when four candidates from the same party ran. Some would even argue that Ralph Nader played a similar spoiler role in 2000, although he won a much smaller share of the vote.
11 Noel: Ten Things Political Scientists Know that You Don t 9 On the other hand, we avoid breakdowns pretty well. We do it with institutions, like political parties and party primaries. When parties narrow our choices, they are doing us a favor. Indeed, political scientist E.E. Schattschneider (1942) argues that democracy is unthinkable save in terms of parties. We also do so with ideology. If most political questions can be reduced to liberal and conservative options even if everyone is on a spectrum from left to right, with many moderates then that alone might be enough to escape Condorcet s and Arrow s problems (Black, 1958, Downs, 1957, Hotelling, 1929). #4. There Is No Such Thing As A Mandate Take items #1, #2, and #3 together, and it is hard to interpret elections the way that politicians and pundits want us to. Economic fundamentals guide voters who might not have well-defined attitudes to vote in a system that cannot satisfy all the demands of democratic decision-making. This is not a formula for sending a clear message to anyone. In a narrow sense, winning the election gives you a mandate. You are now legally empowered to exercise your authority. But after every election, pundits will declare a more complex mandate. They will tell you not simply who won, but what that victory means. The winners won because they promised X and voters wanted Y and were afraid of Z. These narratives are created after the fact by people who want you to think one thing or another (Grossback, Peterson, and Stimson, 2006, 2007). Winners claim a mandate to change everything, and losers explain it all away as an anomaly. But exit polls saying that some voters cared about some things are thin reeds on which to spin out the will of the people. #5. Duverger: It s The Law Social scientists are notoriously unwilling to declare anything with certainty. Physical science is full of laws; we just have findings. Except for Duverger s Law. It was put into print in the 1950s by Maurice Duverger but understood for much longer (Riker, 1982). To wit: The simple-majority single-ballot system favours the two-party system. (Duverger, 1951, 217). That is, with an electoral system like ours, we are likely to get a two-party system. Relatedly, Duverger notes that proportional representation systems tend to support many more parties. As a law, Duverger s claim is clearly false. His predictions fail in many countries. Canada and India have multiparty systems, and Great Britain just elected a hung parliament. But the general tendency does hold up, for two reasons.
12 10 The Forum Vol. 8 , No. 3, Article 12 First, from the point of view of party leaders: Politicians, knowing about the problems with voting mentioned in item #3, do not want a minority opinion to win. They will bow out if they will be spoilers. Even if they believe their party or candidate is best, they might be better off merging with a larger party. Second, from the point of view of voters: If your first choice does not have a chance to win, you might want to vote for a second or third choice who might win. You might otherwise feel like you are wasting your vote, and in a real way, you would be. The Electoral College in the United States tends to exaggerate this effect (Cox, 1997). Almost every state s electoral votes are winner-take-all to the plurality winner in the state. This magnifies the mechanisms in single-member districts. The framers expected a lot of regional candidates for president, with no one winning a majority in the Electoral College. But American parties have, almost from the beginning, tried to appeal across the nation, rather than to regional niches. And so one of the two parties is sure to win the presidency. Recall that in 1992, Ross Perot won almost 20 percent of the vote, but no Electoral College votes. These observations tend to anger advocates of third parties and independent candidates. But noting that the system has these effects is not the same as advocating them. If activists want third-party candidates to succeed, their first step should be to reform the electoral rules. The periodic enthusiasm of journalists for independent candidates, from Anderson and Perot to hypotheticals like Michael Bloomberg or the loser of the latest primary, is not enough. In the meantime, quixotic third party candidacies will be quixotic. Perhaps the most important lesson to draw from Duverger s Law is that voting is not about expressing your opinion. It is about coordinating with other voters. And your institutions determine how you must coordinate. Even in multiparty systems, governments are a compromise among parties to form a majority of in the legislature. In a single-member-district, majority-rule system, the voter has to (or gets to) compromise with fellow voters. The chief difference between two-party systems and multiparty systems is that in a two-party system, the governing coalition is formed informally before elections, while in a multiparty system it is formed formally after elections, in the legislature. #6. Party On It is a fantasy of many a journalist and voter that politicians will come to Washington (or wherever) and sit down and just make good decisions. Politicians in this fantasy will set aside their differences and just find out what is best for the country, and then do it. They will not try to score political points. They will not stick to their ideologies. They will just do what is good for America.
13 Noel: Ten Things Political Scientists Know that You Don t 11 In other words, can t we all just get along? Unfortunately, no, we can t. Policy disagreements happen because people disagree about policy. Liberals believe the government has an important role to play in managing the economy, and conservatives do not. Conservatives believe that the government must protect a set of cultural values that liberals do not share. It is true that politicians also want to win, and scoring political points is a part of that. But this winning is in service to policy goals that are divergent. Some compromises are just incoherent. How important the policy goals of political parties are by comparison to their re-election goals is a matter of some debate in political science. Some emphasize the importance of the career goals of ambitious politicians (Schlesinger, 1994, Aldrich, 1995), while others note that party divisions are driven by policy-demanding activists and organizations (Bawn, Cohen, Karol, Masket, Noel, and Zaller, 2006, Cohen et al., 2008, Masket, 2009). Yet there is wide agreement that even career-minded politicians have to listen to their policyoriented, activist base. The result, of course, is polarization, the dysfunction of the day for political journalism. It is important to put polarization in context. Sixty years ago, the Democratic and Republican parties were hard to distinguish on policy grounds. Some observers complained that the parties were not effective for this reason. A commission put together by the American Political Science Association (American Political Science Association, 1950) concluded that lack of ideological distinction between the two parties prevented the parties from acting responsibly. From this criticism developed the notion of Responsible Party Government. The idea is simple: The major parties campaign on distinct platforms. Once elected, they enact those platforms. Then the voters can see what life is like under those platforms. If things go well, they can re-elect that party. But when what the parties stand for is muddled, it is hard to assign credit and blame. And so, even though no elected official has a mandate (see item #8), they should all enact the policies on which they campaigned. In fact, they should enact them whether they won in a landslide or not. There are a lot of problems with the current American version of Responsible Party Government. Partisan obstruction can prevent majorities from enacting their programs, and parties can capitalize on or be punished for things they had nothing to do with. And parties focused on winning elections are not necessarily motivated to achieve policy goals, just to appear as if they are. Nevertheless, Responsible Party Government is a reasonable standard. If Schattschneider is right that parties are necessary for democracy, then we have to embrace political parties, not resist them.
14 12 The Forum Vol. 8 , No. 3, Article 12 #7. Most Independents Are Closet Partisans Fortunately, most Americans are willing to give parties a chance. You just would not know it from the way independents are revered. The biggest non-story of the last half-century is the rise of political independents. Figure 2: Partisans and Independents, % 40% 35% Weak Partisans 30% 25% Strong Partisans 20% Independent "Leaners" 15% 10% True Independents 5% 0% Percent of respondents who self-identify as strong partisans, weak partisans, independents who lean toward one of the two major parties, and true independents. American National Election Studies. It is true that if you ask a survey respondent if they identify with a major party or are independent, a growing number over the last several decades will say they are independent. The problem is that a majority of those independents act like partisans when it comes time to vote or take positions on issues. The American National Election Study asks self-identified independents if they lean toward one party or another. Pretty much all of the growth in independents is among these leaners. (See Figure 2.) And research has demonstrated (Keith, Magleby, Nelson, Orr, Westlye, and Wolfinger, 1992) that in so many ways, the leaners act like partisans. In fact, many act like strong